Beyond 2011 Beyond 2011: Administrative Data Sources Report: NHS Patient Register November 2012 Background The Office for National Statistics is currently taking a fresh look at options for the production of population and small area socio-demographic statistics for England and Wales. The Beyond 2011 Programme has been established to carry out research on the options and to recommend the best way forward to meet future user needs. Beyond 2011 is considering a range of options including census, survey and administrative data solutions. Since ‘census-type’ solutions are already relatively well understood most of the research is focussing on how surveys can be supplemented by better re-use of ‘administrative’ data already collected from the public. The final recommendation, which will be made in 2014, will balance user needs, cost, benefit, statistical quality, and the public acceptability of all of the options. The results will have implications for all population-based statistics in England and Wales and, potentially, for the statistical system as a whole. About this paper This report presents research reviewing the scope and quality of the NHS Patient Register as a potential source of data for use, within the Beyond 2011 Programme, by the Office for National Statistics. The NHS Patient Register contains a record for every person who is registered with an NHS General Practitioner in England and Wales and, although this source of data is not designed to specifically measure populations, it is thought to be one of the largest data sources containing information on virtually everyone in England and Wales. This document is the start of a series of ‘Sources’ reports to be published over coming months. Over the course of the Beyond 2011 Programme, regular reports will be published which will focus on the quality of potential administrative data sources. For more information Search Beyond 2011 @ www.ons.gov.uk or contact : beyond2011@ons.gov.uk © Crown Copyright 2012 Beyond 2011: Administrative Data Sources Report: NHS Patient Register Table of Contents 1 Executive Summary ............................................................................................... 2 2 Introduction ............................................................................................................ 3 3 Background ............................................................................................................ 3 4 Potential Data Source – an overview .................................................................... 4 4.1 The NHSCR ............................................................................................................. 4 4.2 The NHS Patient Register ........................................................................................ 5 4.3 Selection of Data Source.......................................................................................... 5 4.4 Data Collection: The Registration Process ............................................................... 5 4.5 Overview of the Data ONS Receives ........................................................................ 6 4.6 Quality Procedures................................................................................................... 6 5 Overview of the NHS Patient Register Data Extract............................................. 7 5.1 Data Quality ............................................................................................................. 8 5.1.1 Linkability ............................................................................................................... 13 5.2 Sex Ratios in the Extract ........................................................................................ 14 6 Comparison Between NHS Patient Register and 2011 Census Estimates ....... 15 6.1 National Level Comparison .................................................................................... 15 6.2 Local Authority Level Comparison .......................................................................... 23 6.2.1 Local Authorities with a Higher NHS Patient Register Count than the 2011 Census Estimates ............................................................................................................... 26 6.2.2 Local Authorities with a Lower NHS Patient Register Count than the 2011 Census Estimates ............................................................................................................... 27 6.2.3 Local Authorities with a Similar NHS Patient Register Count to the 2011 Census Estimates ............................................................................................................... 27 6.3 Local Authority Examples ....................................................................................... 28 7 Further Considerations and Summary ............................................................... 37 8 Future Changes to the NHS Patient Register ..................................................... 37 9 Conclusions and Recommendation for use in Beyond 2011 ............................ 38 References .................................................................................................................................. 40 Appendix A: NHS Family Doctor Services Registration Form (GSM1)................................... 41 Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 1 1 Executive Summary This report presents research reviewing the scope and quality of the NHS Patient Register as a potential source of data for use, within the Beyond 2011 Programme, by the Office for National Statistics (ONS). The NHS Patient Register contains a record for every person who is registered with an NHS General Practitioner (GP) in England and Wales and, although this source of data is not designed to specifically measure populations, it is thought to be one of the largest data sources containing information on virtually everyone in England and Wales. However, the number of patients on the register has exceeded ONS’s population estimates in every year since 1961 indicating that there are ‘coverage’ issues that will need to be overcome in order for the dataset to be of maximum use. There are a number of reasons why the Patient Register is higher than the population estimate, but this over coverage at national level also hides pockets of under coverage. Over coverage at the national level is most likely to result from people who emigrate from the UK, but do not tell their GP and their record lies dormant for some time and duplicate records. In terms of under coverage the Patient Register does not include Armed Forces personnel who have their own medical system. There are also issues relating to the delay between people moving and changing their GP registration that result in over coverage in the area they have left and under coverage in the area they have moved to. The main findings from comparing the NHS Patient register (reference date of 23 March 2011) with 2011 Census data (reference date of 27 March 2011) are: • at the national level the Patient Register exceeds the Census 2011 estimate by 4.3 per cent with the bulk of this difference found for people aged between 20 and 64 • the NHS Patient Register sex ratio, (males to females), exceeds the census sex ratio for people aged 27 to 68 years • at the local authority level the NHS Patient Register count is within three per cent of the census estimates in 41 per cent of local authorities. 75 per cent of local authorities were within five per cent of the census estimates and 97 per cent of local authorities were within ten per cent of the census estimates • there is considerable variation in the results when the data are compared by local authority, age and sex, highlighting that the overall difference comprises a number of different issues which impact in different ways locally Whilst there are a number of issues with the NHS Patient Register as a source of data to support Beyond 2011 purposes, it will be an important source for the Programme because it: • has a very high level of population coverage (based on a residential base). • has good quality data with very low levels of missing values for key variables • has a unique identifier (NHS number) allowing for anonymised matching with itself over time Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 2 • can be obtained for a comparable reference period. • is produced in a comparable way across all areas of England and Wales, albeit with different processes in both countries The NHS has changed its administrative structure over time and will do so again in the future. ONS will continue to assess the potential impact any such changes might have on the content and availability of the Patient Register. In summary, the NHS Patient Register provides broad coverage of people within England and Wales, and should prove to be an important source of data for the Beyond 2011 Programme. 2 Introduction The principle of Beyond 2011 is relatively simple – the Programme is investigating the best way of producing the population and small area socio-demographic statistics needed to support the effective administration of the country. We are carrying out a programme of research looking at all of the possible approaches to producing this data, then assessing each of these against a clear and agreed set of criteria in order to help us decide on the best way forward. This document, (S1): Administrative Data Sources Report (Patient Register), is the first in a series of detailed assessments setting out the results of reviews of individual sources. We anticipate that further reports will be published including:S2: Administrative Data Sources Report (Household Electoral Registration) S3: Administrative Data Sources Report (School Census for England and School Census for Wales) S4: Administrative Data Sources Report (Higher Education Student Data); and S5: Administrative Data Sources Report (Department for Work and Pensions/HM Revenue and Customs Customer Information System - initial assessment) In making use of data from administrative sources attention must be given to understanding the processes and procedures associated with the collection, collation, processing and validation of the information and the implications these processes may have for the underlying quality of the data. In particular, it is important to take account of the differences between data collected for administrative and statistical purposes and, where necessary, to make allowances for differences in data definitions and classifications as well as variations in timeliness and reference points. This report brings together this information with a quantitative comparison of the Patient Register and the 2011 Census to inform the use of this dataset within the Beyond 2011 Programme. 3 Background All of the indications are that the census held in 2011 has been highly successful – but there are clear signs that taking the census is becoming increasingly challenging and costly. The dynamic nature of populations, advances in information technology and demand for more frequent and more detailed statistics are driving changes in methods. This trend can be seen across many developed countries. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 3 ONS set up the Beyond 2011 Programme in April 2011 to take a fresh look at options for meeting future user needs for population and small area socio-demographic statistics. The Beyond 2011 Programme is studying a range of statistical options including: • • • census-type solutions administrative data solutions survey solutions Since census type solutions are already relatively well understood the majority of the Programme’s research work is focussed on investigating ways of making more use of existing administrative datasets combined with targeted surveys. Recommendations on the way forward, to be made in 2014, will be informed not only by the statistical viability of the potential solutions, but also by a full understanding of user requirements, public burden, costs and public acceptability. Evaluating the quality of administrative data sources that may be used within the Beyond 2011 programme is a key element of the research. The outcomes will inform the benefits and challenges associated with using these data sources. The NHS Patient Register is one such administrative source. ONS has access to an extract of demographic information from the NHS Patient Register. It contains a record for every person who is registered with a GP in England and Wales. Due to its broad coverage of persons in England and Wales, the NHS Patient Register could prove to be a key source of data for the Beyond 2011 Programme. 4 Potential data source – an overview There are two administrative sources available to ONS dealing with persons registered for medical services. These are; 4.1 • the NHS Central Register, (NHSCR), a centralised system for patient administration, and • the National Health Service (NHS) Patient Register, a detailed list of patients with additional geographical details. Primary Care Trusts (PCTs) are currently responsible for maintaining the NHS Patient Register in England with the NHS Wales Shared Services Partnership responsible for Wales. The NHSCR The NHSCR provides a comprehensive system to assist with NHS patient administration in England and Wales. Responsibility for the register transferred from ONS to the NHS Information Centre (NHSIC) on April 1st 2008 (NHS 2009). The register contains broad, but not complete, coverage of the population of England and Wales. Updates occur as a result of weekly submissions of new births and deaths from the Registrars of Births and Deaths, and for new registrants with GPs (such as migrants). It is also routinely updated with details of name changes, immigration, enlistments to Armed Forces, adoptions, entries into service medical officer care and cancer registrations. The Administrative Data Liaison Service estimate approximately 9 million changes are made to the NHSCR each year, including the creation of around 600,000 new records per year (Administrative Data Liaison Service 2012). The work of the NHSCR covers a range of services, including recording the transfer of patients between health authorities. Each record in the extract of the register provided to ONS contains the patient’s NHS number, forename and surname, date of birth and date of acceptance by the health authority. As of 2006, PCTs replaced health authorities as the main health geography in England. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 4 4.2 The NHS Patient Register The coverage of the NHS Patient Register is all persons registered with a GP in England and Wales 1. The NHS Patient Register is used to maintain an accurate list of all persons registered with a GP, allowing the timely transfer of medical records and correct payments to doctors. It contains a list of everyone who is, or ever has been, registered with a GP in England and Wales since the NHS was founded in July 1948 (NHS 2009). Although this source of data has a specific administrative purpose and is not designed to specifically measure the population, it is thought to be one the largest data sources containing information on virtually everyone in England and Wales. The NHS Patient Register extract provided to ONS currently contains approximately 58 million records, extracted from 87 separate National Health Applications and Infrastructure Services (NHAIS) sites; 82 of which are in England and 5 of which are in Wales. Every NHAIS site holds a register of the patients registered with GPs within their area of responsibility. The register contains a number of variables including the NHS number, gender, date of birth, date of acceptance at the health authority and unit postcode of the address of each patient. It does not contain any information about the health of patients. 4.3 Selection of data source The NHS Patient Register is a complete list of all persons registered with a GP in England and Wales whereas the NHSCR extract currently available to ONS relates only to changes which have taken place over the year, e.g. an address change or a change in GP, rather than a complete register of all patients. It has therefore been decided to focus on the NHS Patient Register as it is more suitable for Beyond 2011 purposes, in our research to estimate the size of the population. 4.4 Data collection: the registration process The registration process begins when a patient attends a GP surgery and asks to register with a GP. The patient is usually asked for their medical card and for their date of birth; gender; current address and postcode; name of their last GP; and previous address and postcode. The patient is also asked whether they have come from abroad. If the patient is accepted onto the register (the default is to accept), the registration will proceed by the information being sent to the PCT (for England) or the NHS Wales Shared Services Partnership (for Wales) either on paper via NHS form GMS1 (see Appendix A) or electronically via a system known as GP Links. At regular intervals - usually weekly - GP practices notify the PCTs (for England) or the NHS Wales Shared Services Partnership (for Wales) of new patients. These are processed by the PCT or the NHS Wales Shared Services Partnership with the patient records being checked to see whether the person is already present on the local database. If the person is new to the area, NHSCR are informed so that medical records can be obtained. The NHS number is intended to be unique. There are, however, a small number of cases where the NHS number is duplicated across records (See section 5.1 below). 1 Whilst responsibility for the NHS in Wales is devolved, the information is provided to ONS along with information for England by NHS Connecting for Health Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 5 4.5 Overview of the data ONS receives ONS obtains a copy of PCT registers on an annual basis 2 and combines these to create an integrated dataset for England and Wales. ONS uses the Patient Register dataset to measure internal migration; as part of the calculation of the mid-year population estimates; and within the ONS longitudinal study 3. Patient Register data were also used as part of the process of quality assuring the 2011 Census results. For internal migration purposes downloads are taken on 31st July to enable migration estimates to be made for the year ending 30th June, allowing a months delay for registrations following moves before 30th June. Agreement was reached with the NHSIC for additional ‘Enhanced’ NHS Patient Register (EPR) extracts in June 2010. The EPR extract contains the following additional variables: surname; forename; other forenames; address – up to five separate lines. These additional variables make it possible for ONS to perform linkage with completely ‘anonymised’ data between the EPR and other administrative data sources (see section 5.1). Analyses in the present report were performed using the EPR dataset which was supplied to ONS on 23 April 2011. 4.6 Quality procedures Some quality checks are carried out by the NHS Connecting for Health (CfH) prior to releasing an extract to ONS. Details on these are given below. i) Identification of temporary records Temporary records are created for NHS patients receiving treatment in an area that is not their usual residence. They can be identified because they are issued with a temporary NHS number. These patients can appear more than once on the NHS Patient Register and so records with a temporary NHS number are removed from the register before any further processing is done. ii) Identification of records with incomplete data As part of the validation process, records are identified which fail a basic range or validation check. Records that fail basic checks are written to a separate file and are referred to as incomplete records. iii) National Duplicate Registration Initiative (Audit Commission) The Audit Commission periodically carry out a data cleaning exercise. National Duplicate Registration Initiatives (NDRI) have previously been undertaken in 1999, 2004 and 2009 and are the only occasions when the full NHS Patient Register is subjected to a de-duplication process(Audit Commission 2012). The prime reason for these exercises is to address list inflation. List inflation is a financial concern for the NHS as the number of people on the NHS Patient Register in any given area determines the amount of funding provided to GPs in that area. Thus, for every erroneous record on the NHS Patient Register, a set fee will be paid to the GP site to which that record is associated. The NDRI carried out in 2004 resulted in almost 2 3 ONS has had legal access to NHS Patient Register data since 1999. See ONS Longitudinal Study for more details Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 6 185,000 patient records being removed from the patient register. These deleted records were categorised as follows: duplicate records (34 per cent); deceased persons (22 per cent); ‘gone aways’ (persons identified as having moved out of the area) (28 per cent); age reports (eight per cent); temporary NHS numbers (seven per cent); removed asylum seekers (one per cent), (Audit Commission 2006). For the most recent exercise in 2009/10 approximately 58 million patient records were extracted from the NHAIS system. The NDRI 2009/10 produced more than 750,000 data matches of which 95,000 records were removed; representing 0.16 per cent of the total number of records extracted for the exercise. The deleted records were categorised as follows 4: Deceased persons (34 per cent); Duplicate records (31 per cent); multiple occupancy (ten or more at an address) (21 per cent); removed asylum seekers (ten per cent); Age reports (four per cent), (Audit Commission 2012). While not all deleted record categories can be directly compared between NDRI 2004 and 2009, the Audit Commission (2012) still concluded that the reduction in the total number of deleted records, (from 185,000 in NDRI 2004, to 95,000 in NDRI 2009), is evidence that patient register data quality has improved between 2004 and 2009. 4 Note ‘temporary NHS number’ and ‘gone away’ matching was not undertaken in NDRI 2009 whereas ‘multiple occupancy matching’ was introduced only in NDRI 2009 (Audit Commission 2012). Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 7 5 Overview of the NHS Patient Register data extract Figure 1 shows the number of records in the April 2011 NHS Patient Register dataset for persons aged 0-84 with notable features being: (i) an upward spike in frequency for persons aged 61-63 years which corresponds to the baby-boom in the years immediately after World War Two (1946-1948); a dip in frequency for those aged 32-38 reflecting the lower birth rate between 1973 and 1979 and; the increase in births in the last decade is seen for those aged 0-10. (ii) (iii) Figure 1. 2011 NHS Patient Register of England & Wales by total persons 1000 900 Population (thousands) 800 700 600 500 400 300 200 100 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 Age Source: NHS Patient Register 2011. 5.1 Data quality Once ONS receive the data a number of quality checks are carried out. The following section details some checks and the resulting implications for the Beyond 2011 Programme where applicable. Duplication The last NDRI was carried out in 2009/10 (Audit Commission 2012). Despite this, ONS needed to remove a number of duplicates which existed in the 2011 EPR data and thus it would appear that some duplication is still present in the data extract. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 8 ONS collated the 87 supplied files which made up the April 2011 EPR extract, resulting in a total of 58,509,255 records. Of these, 22,122 duplicate records were identified by de-duplicating on the supplied NHS number unique identifier and retaining the most recently updated record. These 22,122 duplicate records were then examined to identify those where the year of birth differed from that recorded on the retained record. 364 were found to fulfil this criterion, and of these 41 duplicate NHS numbers were found which referred to a different record than the retained record. These records were reinstated in the data set. The remaining 323 records in the set could reasonably be judged to refer to the same people, although these records did include minor discrepancies in the date of birth variable. The final total was 58,487,174. Table 1 summarises this information. Table 1: Duplicate NHS Number analysis Record count 58,509,255 22,122 364 Initial record count Duplicate NHS Numbers found Duplicate NHS Numbers where associated dates of birth do not match Duplicate NHS Numbers that do not refer to same 41 person* * Records were classified as not belonging to the same person where the recorded name was different or there were minor differences between names and significant ones between dates of birth. Source: NHS Patient Register 2011. Missing data Table 2 shows the percentages of missing/non-missing data, in the April 2011 EPR extract, for each of the twelve variables which are considered as potentially key for Beyond 2011 purposes. As Table 2 indicates, there are no missing data for NHS numbers; Gender; Surname; Forename and the fourth address line. Further there are just 2 missing values for date of birth. A small number of postcodes are missing but this represents less than 0.01 per cent of the total records in the NHS Patient Register. The largest instances of missing data are amongst four of the five address fields (range 2.1 per cent - 80.5 per cent missing data). There is no missing data for the Address4 variable and this is likely to be because the fourth address line typically contains the name of the postal town or city in which the patient is resident and this is a mandatory field when creating a record in the Patient Register. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 9 Table 2. April 2011 Extract - Enhanced Patient Register Key Variable completeness Variable name Variable type Non-missing April 2011 Missing April 2011 (per cent) (per cent) 58,487,172 2 DateOfBirth Numeric (99.99) (0.01) 58,487,174 0 NHSNumber Character (100.0) (0.0) 58,487,174 0 Gender Character (100.0) (0.0) 58,487,174 0 Surname Character (100.0) (0.0) 58,487,174 0 Forename Character (100.0) (0.0) 8,672 58,478,502 Postcode Character (99.99) (0.01) 11,397,125 47,090,049 Address1 Character (19.49) (80.51) 57,268,749 1,218,425 Address2 Character (97.92) (2.08) 30,965,608 27,521,566 Address3 Character (52.94) (47.06) 58,487,174 0 Address4 Character (100.0) (0.0) 16,863,351 41,623,823 Address5 Character (71.17) (28.83) Source: NHS Patient Register 2011. Coverage issues There is evidence that the NHS Patient Register is subject to both over coverage and under coverage. In April 2011 the NHS Patient Register record count (58,471,500) was 4.3 per cent greater than the 2011 Census population estimate (56,075,900) of England and Wales (ONS 2012a). Similarly, the number of patients on the registers has exceeded the number of people resident in the country when compared to ONS Mid-Year Estimates, in every year since 1961 (Scott & Kilbey, 1999). Possible reasons for over coverage are that some patients are registered in more than one area; some patients may have more than one NHS number and finally that some patients remain on the NHS Patient Register after leaving the country and/or having died (Audit Commission 2012). Additionally some people are eligible to register with a GP that would not be included in the ONS definition used for producing the population estimates. Some of the main reasons for coverage issues can be categorised as follows: • multiple area registrations: the current registration system does not require a ‘new’ patient to provide proof of identification to support their registration. Therefore, the possibility exists for successful applications being made by a single person, but at different GP sites. For example, a person who is registered with a GP in Hampshire permanently moves to a house in Lancashire and registers with a GP near their new residence. However, when registering they provide information which is sufficient to allow them to register with their new GP, but insufficient to allow the GP site to identify the person as being already on the NHS Patient Register, e.g. due to the person spelling their name differently to how it appears on the NHS Patient Register and/or by providing a date of birth Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 10 in an ambiguous date format DD-MM-YY / MM-DD-YY; (e.g. 01-03-73 versus 03-01-73). In this instance it is possible that a new record, with a new unique NHS number could be created, meaning this person would then have two separate records, with two different NHS numbers, on the NHS Patient Register 5 6 • duplicate NHS number: records which have a duplicate NHS number can occur in one of two ways. The first type of duplicate is two or more identical records, the majority of which are thought to occur temporarily, at any given point in time, where a person is in the process of transferring from one GP to another (Audit Commission 2012). The second type, the cause of which is unclear, is where the records of two different people have the same NHS number. According to the Audit commission this applies to a small number of records but was highlighted as a data quality issue (Audit Commission 2012) • lags in the recording of births, deaths and migrants on the NHS Patient Register 5: potential time delays in notification or no notification of events (births, deaths, internal and international moves) can lead to patients remaining on registers when they should be removed or in the case of births or immigration, not appearing on the list straight away. Below are some details on different event notification procedures o births: NHSCR are responsible for distributing NHS numbers to the Registrars of Births and Deaths (RBDs), who, in turn, allocate them as each new birth is registered. Sometimes the RBDs report the births directly to the health authority, but usually it is the parent who registers the child with a GP. As some parents may take longer to do this than others there may be a lag where a child may not appear on the register for a number of weeks or months after being born. Further, if a child is registered solely for private healthcare they will not appear on the NHS Patient Register at all resulting in under coverage. o deaths: currently NHAIS sites receive notifications of patient deaths from the Personal Demographics Service National Back Office, which links the General Register Office register of recorded deaths to patient lists (Audit Commission 2012). However, some NHAIS sites have expressed concern that they do not have a system in place that checks for patients who have died whilst abroad. As a result, some persons who die whilst abroad may not be removed from the NHS Patient Register and cause over coverage (Audit Commission 2012). o embarkations: NHSCR receives details of embarkations 6 from people handing in their medical cards and from the Department of Work and Pensions, Home Office and Immigration officers. However, if somebody does not inform the authorities that they have emigrated, or intend to emigrate, it is likely that they will remain on the NHS Patient Register after leaving the UK, leading to over coverage. There is also a clear incentive for emigrants to remain on the NHS Patient Register as it allows them to access health care services, free of charge, should they return temporarily to the UK; e.g. to visit relatives. o international migrants: international migrants may not register with a GP when they arrive in England or Wales. Some may not register at all. Further discussion on this can be found in the definitional differences section below. See Smallwood and Lynch (2010) for more details on the issue of lags. People leaving the country. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 11 • geographical variations in data quality: NHS systems are administered on a regional basis and inevitably there is variation in quality between regions. This is also true at the country level as the NHS Patient Register is administered by different bodies for England and Wales. This could cause small differences in data quality between the two countries. In addition the NHS is devolved in Wales, which may result in long term risk of any different changes in policy, management or processes, which could lead to future incomparability between England and Wales. Even at the local level there is likely to be variation between GP sites in efficiency of record keeping. Data quality initiatives, such as identifying multiple registrations and cancelling student registrations, implemented by NHS Information Systems also vary geographically, for example in terms of the regularity with which data quality programmes are implemented. This is likely due to the availability of resources, e.g. financial, available personnel and technical expertise, differing from one area to another (Audit Commission 2012). • definition differences - definition of usually resident: according to ONS (2009) a usual resident of the UK (for census output purposes only) is anyone who, on 27 March 2011 was (i) in the UK and has stayed or intended to stay in the UK for a period of 12 months or more or; (ii) had a permanent UK address and was outside the UK and intended to be outside the UK for less than 12 months. There are many reasons why a person may appear on the NHS Patient Register but not be deemed as usually resident, according to the census definition. For example, a person may register with a GP, and thus appear on the NHS Patient Register irrespective of whether they intend to stay in the UK for at least 12 months. The relationship between the population covered in the NHS Patient Register and the usual resident population is illustrated in Figure 2. Here, we see that for a specific area at a specific time, the population covered in the NHS Patient Register may differ from the usual resident population as a result of: the inclusion of persons staying in England and Wales for less than 12 months; the omission of persons who are not registered with a GP; people remaining registered after they have emigrated or died; erroneous list cleaning (removing patients when they have not moved); issuing a new NHS number to someone who is already on the list and lags in international migrants registering with a GP. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 12 Figure 2. Relating the NHS Patient Register to the Usual Resident Population in Area j at Time t Failure to de-register or registration lag for out-migrants Linkability The NHS number is unique enough to allow longitudinal linking of the NHS Patient Register with itself. However, as NHS numbers are not included on any other administrative data source available to ONS, linkage between administrative sources using NHS number will not be possible. Despite this, linkage may still be possible by using combinations of other variables such as surname, date of birth and address information. It is recognised that this raises concerns relating to the privacy of data about individuals and households and that steps need to be taken to ensure confidentiality of any data used. For this reason, Beyond 2011 is currently developing methods to enable linkage with completely ‘anonymised’ data through a number of pre-processing steps. The results of research to date are very promising and have been applied to current research. We will be discussing our approach with experts over the next few months and will publish more details of the methods in 2013. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 13 5.2 Sex ratios in the extract Figure 3 shows the ratio of men to women on the NHS Patient Register for years 2001, 2006 and 2011. Sex ratios are often used as a quality measure of data by age and sex as they are independent of the absolute numbers of males and females in a large population (Smallwood & De Broe 2009). Figure 3 shows the sex ratios at the younger ages as fairly consistent and as expected from a large Western developed country (around 105 boys per 100 girls). However from the age of 18 a dip is seen in the sex ratio which peaks around the age of 23. This dip in the sex ratio is apparent in the three years studied (2001, 2006 & 2011) and thus it can be concluded that the dip is not ageing forwards. From this dip the sex ratio rises to a bulging profile from age 30 to 65 years. This dip in the national sex ratio is primarily determined by international migration. Previous research has focused on the trends in the NHS Patient Register sex ratio for this specific age group and found the sex ratio to be driven primarily by the registration and de-registration of female migrants (Smallwood & De Broe 2009). An increase in female migrants registering at specific age groups could cause a lower sex ratio for these same age groups. In contrast, the de-registration of female migrants, at other age groups, could produce a higher sex ratio for these same age groups. Previous research conducted by ONS found females to be more compliant with administrative requirements and to register with a GP on arrival at the new location (Smallwood & De Broe 2009). The ‘bulging’ profile in the NHS Patient Register above the age of 30 may be the result of delays in male migrants registering at a GP practice and/or male emigrants not being deregistered when they leave the UK. Figure 3. Sex Ratios of the NHS Patient Register for England & Wales for years 2001, 2006 & 2011 110 Ratio (number of men per 100 women) 100 90 80 70 60 50 40 2001 30 2006 20 2011 10 0 0 5 10 15 20 25 30 35 40 45 Age 50 55 60 65 70 75 80 Source: NHS Patient Register 2001, 2006, 2011. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 14 6 Comparison between NHS Patient Register and 2011 Census estimates In this section the NHS Patient Register count is compared with the 2011 Census population estimates at both the national and local authority level. The following analysis looks at the comparison of the 2011 Census population estimates (reference date of 27 March 2011) and the 2011 NHS Patient Register count (reference date of 23rd April 2011). 6.1 National level comparison Total population At the England and Wales national level, the 2011 NHS Patient Register count is greater than the 2011 Census estimates by 4.3 per cent (ONS 2012a). Figure 4. 2011 NHS Patient Register count versus 2011 Census estimates for England & Wales by total persons 1000 900 800 Population (thousands) 700 600 500 400 300 Patient Register 200 Census 100 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 Age Source: NHS Patient Register 2011, Census 2011 Figure 4 shows a comparison of the age distribution of the two sources. It can be seen that there is a good correspondence from nought to 20 years. From 20 years onwards, the NHS Patient Register count exceeds that of the 2011 Census estimates, with a bulging pattern between the ages 20 and 64. At this point the lines meet again. This pattern supports the suggestion presented Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 15 in the first part of this report that the NHS Patient Register is subject to net over coverage and therefore a greater count is seen on the NHS Patient Register than that of the resident population. When comparing the two sources at the England national level (Figure 5), the 2011 NHS Patient Register count is greater than the 2011 Census estimates by 4.3 per cent (ONS, 2012a). The age distribution, for both sources follow a similar pattern to that seen at the England and Wales national level. Figure 5. 2011 NHS Patient Register count versus 2011 Census estimates for England by total persons 1000 900 Population (thousands) 800 700 600 500 400 Patient Register 300 Census 200 100 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 Age Source: NHS Patient Register 2011, Census 2011. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 16 When comparing the two sources at the Wales national level (Figure 6), the NHS Patient Register count is greater than the 2011 Census estimates by 3.1 per cent (ONS 2012a). The differences between the two sources follow a similar pattern to that seen for England, although the differences between the two sources are smaller than those seen for England. This could suggest that the Patient Register in Wales experiences less net over coverage than the Patient Register for England. Figure 6. 2011 NHS Patient Register count versus 2011 Census estimates for Wales by total persons 50 Population (thousands) 40 30 20 Patient Register 10 Census 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 Age Source: NHS Patient Register 2011, Census 2011 Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 17 Figure 7 shows a comparison between the 2011 NHS Patient Register and the 2011 Census estimates for men in England and Wales. It can be seen that between the ages of 0 and 21, the two sources are very similar. However, between the ages of 22 and 63, the NHS Patient Register has a higher count of men compared to the 2011 Census estimate. Lastly, between ages 64 and 84 the two sources are again very similar. Figure 7. 2011 NHS Patient Register versus 2011 Census estimates for England and Wales by males 500 450 Population (thousands) 400 350 300 250 200 150 Patient Register Males 100 Census Males 50 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 Age Source: NHS Patient Register 2011, Census 2011 Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 18 Figure 8 shows a comparison between the 2011 NHS Patient Register count and the 2011 Census estimates for women in England and Wales. It can be seen that between the ages of 0 and 19, the two sources are very similar. However, between the ages of 20 and 48, the NHS Patient Register has a higher count of women compared to the 2011 Census estimate. Lastly, between ages 49 and 84 the two sources are again very close. Figure 8. 2011 NHS Patient Register count versus 2011 Census estimates for England and Wales by females 500 450 Population (thousands) 400 350 300 250 200 150 Patient Register Females 100 Census Females 50 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 Age Source: NHS Patient Register 2011, Census 2011. Comparing Figures 7 and 8, it is apparent that there is a greater difference between the two sources for men than for women and that this difference is spread out over a wider age range for men (22-63 years) than for women (20-48 years). This difference can be seen at both the England national level and the Wales national level. Section 5 of this report examined in detail some of the reasons why the 2011 Census estimates and the NHS Patient Register counts are likely to differ. Two of the reasons identified those of multiple registrations and lag in registration, are likely to go some way towards explaining why there are larger differences for men than for women. In particular, the larger difference between the two sources accounted for by men could be a result of men being slower to register/re-register with a GP than women, and men being less likely to de-register when they leave the country. These differences will be important when considering how the NHS Patient Register can be used within the context of Beyond 2011. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 19 Sex ratios at the national level Figure 9 compares the sex ratios in the 2011 NHS Patient Register against those of the 2011 Census estimates for England and Wales. The sex ratios for both sources follow a similar pattern for the ages 0 to 19. At this point 2011 Census estimates show a slightly higher sex ratio up to the age of 26. At 27 years the lines cross and the 2011 NHS Patient Register sex ratio exceeds the 2011 Census sex ratio with a bulging pattern up to the age of 68. From age 68 onwards the sex ratios show a similar pattern again. This variation in the sex ratio for the NHS Patient Register is likely to be caused by international migration. The dip in the NHS Patient Register for the ages 18 to 24 is likely to be due to more female migrants than male migrants registering with a GP when they arrive in the country. However, as these migrants leave the country, if it is the case that female migrants are more likely to deregister, (as discussed earlier), but the male migrants remain on the register; this would result in the pattern seen from the age of 27 onwards. This is compounded by a higher de-registration rate for UK national women who emigrate as compared with UK national men. This will cause over coverage in the NHS Patient Register for some age groups. The census sex ratios show a more consistent pattern across its age groups as would normally be expected, (Smallwood & De Broe 2009). Figure 9. Sex ratios from 2011 NHS Patient Register & 2011 Census estimates for England & Wales 110 Ratio (number of men per 100 women) 100 90 80 70 60 50 40 Patient Register 30 Census 20 10 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 Age Source: NHS Patient Register 2011, Census 2011 Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 20 Figure 10 compares the sex ratios in the 2011 NHS Patient Register against those of the 2011 Census estimates for England. The sex ratios for both sources follow a similar pattern to that at the England and Wales national level. Figure 10. Sex ratios from 2011 NHS Patient Register & 2011 Census estimates for England 110 Ratio (number of men per 100 women) 100 90 80 70 60 50 40 Patient Register 30 Census 20 10 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 Age Source: NHS Patient Register 2011, Census 2011. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 21 Figure 11 compares the sex ratios in the 2011 NHS Patient Register against those of the 2011 Census estimates for Wales. Overall, the sex ratios for both sources follow a similar pattern to that at the England and Wales national level. However the dip at the 18-24 year olds is not as distinct as that at the England national level. The bulging pattern that is seen in the England sex ratios from the ages of 27 to 68 is still seen in the Wales national graph but again is less prominent than that at the England level. Figure 11 Sex Ratios from 2011 NHS Patient Register & 2011 Census Estimates for Wales 110 Ratio (number of men per 100 women) 100 90 80 70 60 50 40 30 Patient Register 20 Census 10 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 Age Source: NHS Patient Register 2011, Census 2011. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 22 6.2 Local Authority Level Comparison When comparing the NHS Patient Register count with the 2011 Census estimates at the local authority level, (Figure 12), the NHS Patient Register count is within three per cent of the census estimates in 41 per cent of local authorities in England and Wales. 75 per cent were within five per cent of the census estimates and 97 per cent were within ten per cent of the census estimates. For those local authorities who had a difference between the two sources of greater than three per cent, analysis shows that the majority of these local authorities had a NHS Patient Register count which was higher than the census estimates. Figure 12. Percentages of Local Authorities in the 2011 NHS Patient Register counts, presented as a percentage difference to Census estimates 45 40 35 Percentage 30 25 20 15 10 5 0 Under 10 to 19.9% Under 5 to 9.9% Under 3 to 4.9% Within 3% Over 3.4.9% Over 5-5.9% Over 1019.9% Source: NHS Patient Register 2011, Census 2011 Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 23 Map 1 shows the comparison between the 2011 NHS Patient Register count and the 2011 Census estimates, by local authority, across England and Wales. Differences between the NHS Patient Register count and the 2011 Census estimates are much more variable across the local authorities and therefore it can be concluded that this variation is dependent on particular local circumstances. Map 1 highlights the difference between England and Wales when comparing the NHS Patient Register count with the 2011 Census estimates. When comparing the two sources at the local authority level, the NHS Patient Register count is within three per cent of the census estimates in 40 per cent of local authorities in England and 59 per cent of local authorities in Wales. This pattern supports the suggestion presented in section 6.1 that the NHS Patient Register in England is subject to a higher net over coverage count than the NHS Patient Register in Wales. Forest Heath and Richmondshire, (the two dark brown shaded areas in map 1) show the census being more than ten per cent higher than the NHS Patient Register. Both of these local authorities have a significant Armed Forces population and this is likely to be the reason behind this difference. Armed Forces and the NHS are discussed further in section 6.2.2. Map 1 also shows that the local authorities where the NHS Patient Register count falls within three per cent of the 2011 Census estimates are often in rural areas where the population is likely to be less mobile. These areas are unlikely to experience as much turnover on the NHS Patient Register as urban ones, as the population remains fairly static and are therefore at less risk of coverage issues. Map 1 indicates that the local authorities that experience more than a three per cent difference between the NHS Patient Register count and the census estimates are often in large urban areas. The map shows London, Manchester, Liverpool, Birmingham and Cardiff to be within this category. Possible reasons for this are discussed in the next section. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 24 Map 1. Percentage difference between the 2011 NHS Patient Register count and the 2011 Census estimates by local authorities in England and Wales, for total persons Source: NHS Patient Register 2011, Census 2011 Contains National Statistics data © Crown copyright and database right 2012 Contains Ordnance Survey data © Crown copyright and database right 2012 Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 25 6.2.1 Local authorities with a higher NHS Patient Register count than the 2011 Census estimates The majority of local authorities show higher NHS Patient Register counts than the 2011 Census estimates. Those local authorities which have the largest difference between the two sources are characterised as being large urban areas, with eight London boroughs being within the top ten local authorities with the highest percentage difference (see Figure 13). Figure 13. Top ten local authorities in England and Wales where the 2011 NHS Patient Register count is greater than the 2011 Census estimates 20 18 Percentage difference 16 14 12 10 8 6 4 2 0 Blackpool Ealing Merton Lewisham Wandsworth Newham Brent Oxford Southwark Lambeth Source: NHS Patient Register 2011, Census 2011 High levels of migration (both internal and international) are often seen in large urban areas. High levels of international migration can result in over coverage on the NHS Patient Register as discussed in section 5.1. Obviously in areas where there are higher levels of international migration the risk of this over coverage is far greater than areas of low international migration. In the areas where the NHS Patient Register count is higher than the 2011 Census estimate there are higher proportions of short-term migrants than in the areas where the NHS Patient Register is lower than the census estimates, (ONS 2012b). Oxford is shown to have the third highest positive percentage difference compared to the 2011 Census estimates. Oxford is not a large urban area like London; however, it does experience one of the highest levels of international migration in England and Wales (ONS 2012b). Coverage issues can also occur when a person moves from one local authority to another as they may not register with a new GP for a significant time period. They will also probably have not deregistered from their GP in their previous local authority. This will cause an under count in one local authority and an over count in the other. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 26 6.2.2 Local authorities with a lower NHS Patient Register count than the 2011 Census estimates There are a small number of local authorities which show a lower count in their 2011 NHS Patient Register than in their 2011 Census estimates. Figure 14 shows that the local authorities of Richmondshire and Forest Heath experience the largest negative difference when comparing the 2011 Census estimates and the 2011 NHS Patient Register. Both of these local authorities contain large Armed Forces bases. Hambleton (bordering Richmondshire), St Edmundsbury (bordering Forest Heath), Gosport (Naval) and Wiltshire (Army) also see a negative difference when comparing the two sources. Figure 14. Top ten local authorities in England and Wales where the 2011 NHS Patient Register count is lower than the 2011 Census estimates 0 Percentage difference Wiltshire Tunbridge Wells North Kesteven Hambleton Shropshire Gosport Rutland St Edmundsbury Forest Heath -4 Richmondshire -2 -6 -8 -10 -12 -14 -16 Source: NHS Patient Register 2011, Census 2011. The Armed Forces personnel are counted in the 2011 Census estimates as they are defined as part of the resident population. In contrast, Armed Forces (and in some cases their dependents) are not seen on the NHS Patient Register as they have their own military health services. This exclusion of the Armed Forces personnel from the NHS Patient Register highlights one of the issues of using an administrative source, designed for another purpose, to measure the resident population. Consequently the NHS Patient Register does not have complete coverage of the entire resident population as illustrated in section 5.1. 6.2.3 Local authorities with a similar NHS Patient Register count to the 2011 Census estimates There are a number of local authorities which show some level of accord between the 2011 NHS Patient Register count and the 2011 Census estimates. Without the benefit of record-level data linking it is not clear whether this does show complete agreement between the actual residents of the area or just coincidental numbers, but it is likely to reflect a higher level of agreement in the data than elsewhere. The majority of the areas showing less than a one per cent difference Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 27 between the NHS Patient Register count and the 2011 Census estimates are rural or comparatively sparsely populated areas. 6.3 Local authority examples When comparing the 2011 NHS Patient Register count with the 2011 Census estimates, the extent of the difference between the two sources can be affected by a number of factors, e.g. sex, age, local circumstances. The following maps are shown to highlight some of the differences that exist across the local authorities within England and Wales, by sex and selected age groups. Maps 2 and 3 show the difference between the 2011 NHS Patient Register count and the 2011 Census estimates for men and women respectively aged 20-24. The majority of local authorities show higher counts for the 2011 NHS Patient Register than the 2011 Census estimates, for both men and women, in this age group. There are a small number of local authorities which show the 2011 Census estimates as being higher than the 2011 NHS Patient Register count. This difference is more common for men than it is for women. For men, 22 per cent of local authorities show a higher count for the 2011 Census estimates compared to the 2011 NHS Patient Register. This is in contrast to only nine per cent of local authorities for women. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 28 Map 2. Percentage difference between the 2011 NHS Patient Register and the 2011 Census by local authorities in England and Wales by males aged 20-24 years Source: NHS Patient Register 2011, Census 2011 Contains National Statistics data © Crown copyright and database right 2012 Contains Ordnance Survey data © Crown copyright and database right 2012 Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 29 Map 3. Percentage difference between the 2011 NHS Patient Register and the 2011 Census by local authorities in England and Wales by females aged 20-24 years Source: NHS Patient Register 2011, Census 2011 Contains National Statistics data © Crown copyright and database right 2012 Contains Ordnance Survey data © Crown copyright and database right 2012 Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 30 Maps 4 and 5 show the comparison between the 2011 NHS Patient Register count and the 2011 Census estimates for men and women respectively, aged 40-44. For 40-44 year old men the majority of local authorities show a difference between the two sources as being between three to ten per cent. However, in some urban local authorities a difference of more than ten per cent is seen e.g. some local authorities within Greater London. This pattern is not the same for women, where the majority of local authorities see the difference between the NHS Patient Register and the census estimates as being within three per cent of each other. For women there are only three per cent of local authorities which have a difference of more than ten per cent between the two sources compared to 41 per cent of local authorities for men. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 31 Map 4. Percentage difference between the 2011 NHS Patient Register and the 2011 Census by local authorities in England and Wales by males aged 40-44 years Source: NHS Patient Register 2011, Census 2011 Contains National Statistics data © Crown copyright and database right 2012 Contains Ordnance Survey data © Crown copyright and database right 2012 Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 32 Map 5. Percentage difference between the 2011 NHS Patient Register and the 2011 Census by local authorities in England and Wales by females aged 40-44 years Source: NHS Patient Register 2011, Census 2011 Contains National Statistics data © Crown copyright and database right 2012 Contains Ordnance Survey data © Crown copyright and database right 2012 Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 33 Maps 6 and 7 show the comparison between the 2011 NHS Patient Register count and the 2011 Census estimates for men and women respectively aged 70-74. For both men and women, the majority of local authorities show the difference between the two sources to be within three per cent of each other (71 per cent of local authorities for men and 84 per cent of local authorities for women). This is in contrast to the 20-24 age group where only 17 per cent of local authorities for men and 16 per cent of local authorities for women show a difference between the two sources as being within three per cent of each other. For the 70-74 age group, the local authorities that show a larger difference than three per cent between the two sources are often in urban areas. Both men and women see the majority of the local authorities within Greater London as having a greater difference than three per cent between the two sources. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 34 Map 6. Percentage difference between the 2011 NHS Patient Register and the 2011 Census by local authorities in England and Wales by males aged 70-74 years Source: NHS Patient Register 2011, Census 2011 Contains National Statistics data © Crown copyright and database right 2012 Contains Ordnance Survey data © Crown copyright and database right 2012 Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 35 Map 7. Percentage difference between the 2011 NHS Patient Register and the 2011 Census by local authorities in England and Wales by females aged 70-74 years Source: NHS Patient Register 2011, Census 2011 Contains National Statistics data © Crown copyright and database right 2012 Contains Ordnance Survey data © Crown copyright and database right 2012 Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 36 Further work comparing NHS Patient Register counts to Census 2011 estimates at local authority level has been carried out by ONS as part of the quality assurance process for Census 2011. NHS Patient Register data at local authority level is available on the ONS website on the Local Authority Quality Assurance page. Detailed information about the Census 2011 quality assurance process used by ONS can be obtained on the ONS website on the Quality assurance and methodology page. 7 Further considerations and summary The comparison between the 2011 NHS Patient Register count and the 2011 Census estimates gives us a greater understanding of the quality of the NHS Patient Register. It is evident that the gaps that exist between the two sources vary with age and sex. It is also clear that at the local level patterns are much more variable and depend on particular local circumstances. Higher levels of international migration have been shown to contribute to net over coverage in the Patient Register (Smallwood & De Broe 2009). For areas where the census estimates exceed the NHS Patient Register, a specific population which the NHS Patient Register does not cover, e.g. Armed Forces, is likely to be the reason. In relation to sex differences, it was found that, at a national level, men contributed more toward the 2011 NHS Patient register count exceeding the 2011 Census estimate than women. Further, the age span (22-63 years) during which men on the patient register exceeded 2011 Census estimates was greater than it was for women (20-48 years). The comparison between the 2011 NHS Patient Register count and the 2011 Census estimates can only highlight the difference between the two sources and not what proportion of it is attributable to under coverage and over coverage. Despite only a small number of local authorities being lower than the census it cannot be assumed that the NHS Patient Register only suffers from over coverage in the majority of local authorities. It is highly likely that under coverage is experienced among specific groups of the resident population but that the more prevalent issue of over coverage in fact masks this. Further research using ‘anonymised’ data linking will help us to understand this issue better. Another aspect for consideration is whether some of the patterns observed in sex ratio data can be accounted for by exploring the various NHS screening programmes to determine how these may affect list cleaning differently according to age and gender. This further research will allow ONS to improve its use of the Patient Register data within the Beyond 2011 Programme. 8 Future changes to the NHS Patient Register So far this report has focussed upon the present state of the NHS Patient Register. However, in this section, three known future administrative/structural changes are highlighted which were identified as potentially having implications for the NHS Patient Register and, consequently, for the Beyond 2011 Programme. Personal Demographics Service (PDS) The PDS is a national electronic database of demographic information for persons in England. In time (unlikely to be sooner than 2015) it will integrate with, and ultimately replace, the following NHS services/systems: NHS Strategic Tracing Service; NHS Number for Babies; NHSCR and the NHAIS (‘Purpose and benefits of the PDS’ 2012). There is also an intention to integrate the system into GP practices such that when a person registers with a GP an authorised user can perform a search to immediately attempt to determine the patient’s NHS number. However, ongoing issues Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 37 with the development of this system mean there is currently no fixed date for when the PDS will become the primary source of patient demographic information in England. A similar, but separate, system is currently being developed in Wales and is administered by the NHS Wales Information Service. The likelihood that the PDS will eventually replace the NHS Patient Register may have positive implications for Beyond 2011, in terms of the availability of additional useful variables e.g. first language. Disbanding of Audit Commission The Audit Commission will be disbanded in the next few years, leaving a question mark over where the data matching expertise for future quality initiatives will be sourced (Audit Commission 2012). With respect to Beyond 2011 this may mean future EPR extracts received by ONS may not have been submitted to a recent, full-scale, list de-duplication exercise. Abolition of PCTs PCTs are expected to ‘cease to exist by 2013’ (Department of Health ‘Equity and Excellence: Liberating the NHS’, p.34) and their primary care responsibilities will transfer to NHS Commissioning board or Clinical Commissioning groups. It is too soon to say whether this will have an impact upon data quality of future NHS Patient Register datasets, though, as the possibility cannot be ruled out, this will need investigating when more information becomes available. 9 Conclusions and Recommendation for use in Beyond 2011 The present report has illustrated that the NHS Patient Register is likely to be a key data source for Beyond 2011 as it: • has very high level of population coverage (based on residential base). • provides basic demographic information on age and sex • has good quality data with very low levels of missing values for key variables • includes information on first registration of international migrants (flag 4) • linkage with other administrative data sources is possible using a combination of anonymised key variables; e.g. surname, date of birth and address information • can be obtained for a comparable reference period Comparisons between the NHS Patient Register and the 2011 Census estimates have revealed that for some age-sex groups (e.g. males aged 20-30 years), there are greater differences between the two sources than for other age-sex groups (e.g. males aged 65+); highlighting where other sources and/or a survey may in particular be needed to help estimate the size of the population in Beyond 2011 administrative data based options. Overall, the NHS Patient Register provides very broad coverage of persons within England and Wales and should prove to be an important source of data for the Beyond 2011 Programme. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 38 Glossary EPR Enhanced Patient Register GP General Practitioner NDRI National Duplicate Registration Initiative NHS National Health Service NHAIS National Health Applications and Infrastructure Services NHSCR National Health Service Central Register NHSIC National health Service Information Centre ONS Office for National Statistics PCT Primary Care Trust RBD Registrars of Births and Deaths Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 39 References Administrative Data Liaison Service GP Patient Register Dataset (2012) retrieved 21st August 2012 Audit Commission (2012) National duplicate registration initiative 2009/2010 – National report. Audit Commission (2006) National duplicate registration initiative 2004 – National report. Department of Health (2011) Technical Guidance for the 2012/13 Operating Framework. Downloaded 23rd July 2012 Department of Health (2010), Equity and Excellence: Liberating the NHS. Retrieved 2nd August 2012 National Health Service Family Doctor Services Registration Form (GSM1) National Health Service (2009) ‘NHS Information Centre takes control of NHS Central register’ Retrieved 21st August 2012 Office for National Statistics (2012a) Census first release figures. Retrieved 21st August 2012 Office for National Statistics (2012b) 2011 Census: Number of non-UK short-term residents by sex, local authorities in England and Wales. Retrieved 11th October 2012 Office for National Statistics Longitudinal Study Office for National Statistics Beyond 2011 Reports and Publications Office for National Statistics Local authority quality assurance Office for National Statistics Quality Assurance and Methodology Office for National Statistics (2011) Methodology Note on production of Population Estimates using a Postcode Best Fit Methodology October 2011. Office for National Statistics (2009) Final population definitions for the 2011 Census. Purpose and benefits of the PDS, (2012). Retrieved 30th July 2012 Scott A & Kilbey T (1999) Can patient registers give an improved measure of internal migration in England and Wales?, Population Trends, Summer 1999. Smallwood S & De Broe S (2009) Sex ratio patterns in population estimates, Population Trends. 137, 41-50. Smallwood S & Lynch K (2010) An analysis of patient register data in the longitudinal study – What does it tell us about the quality of the data?, Population Trends 141, 1-19. Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 40 Appendix A: NHS Family Doctor Services Registration Form (GSM1) Beyond 2011 : Administrative Data Sources Report: NHS Patient Register 41